-
Notifications
You must be signed in to change notification settings - Fork 44.6k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Agent loop v2: Planning & Task Management (part 2) #5077
Conversation
* Rename Performance Evaluations to Best Practices * Move specification of response format from system prompt to Agent.construct_base_prompt * Clean up PromptGenerator class
✅ Deploy Preview for auto-gpt-docs canceled.
|
You changed AutoGPT's behaviour. The cassettes have been updated and will be merged to the submodule when this Pull Request gets merged. |
Codecov ReportPatch coverage:
Additional details and impacted files@@ Coverage Diff @@
## master #5077 +/- ##
==========================================
- Coverage 51.93% 51.16% -0.77%
==========================================
Files 117 122 +5
Lines 4987 5302 +315
Branches 671 724 +53
==========================================
+ Hits 2590 2713 +123
- Misses 2199 2387 +188
- Partials 198 202 +4
☔ View full report in Codecov by Sentry. |
This PR exceeds the recommended size of 500 lines. Please make sure you are NOT addressing multiple issues with one PR. |
You changed AutoGPT's behaviour. The cassettes have been updated and will be merged to the submodule when this Pull Request gets merged. |
This PR exceeds the recommended size of 500 lines. Please make sure you are NOT addressing multiple issues with one PR. |
This pull request has conflicts with the base branch, please resolve those so we can evaluate the pull request. |
This PR exceeds the recommended size of 500 lines. Please make sure you are NOT addressing multiple issues with one PR. |
Conflicts have been resolved! 🎉 A maintainer will review the pull request shortly. |
This PR exceeds the recommended size of 500 lines. Please make sure you are NOT addressing multiple issues with one PR. |
This PR exceeds the recommended size of 500 lines. Please make sure you are NOT addressing multiple issues with one PR. |
This PR exceeds the recommended size of 500 lines. Please make sure you are NOT addressing multiple issues with one PR. |
This PR exceeds the recommended size of 500 lines. Please make sure you are NOT addressing multiple issues with one PR. |
This PR exceeds the recommended size of 500 lines. Please make sure you are NOT addressing multiple issues with one PR. |
This PR exceeds the recommended size of 500 lines. Please make sure you are NOT addressing multiple issues with one PR. |
This has been tested to considerably improve the performance of Auto-GPT on the benchmark, and it contains a number of building blocks for further improvements. |
You changed AutoGPT's behaviour. The cassettes have been updated and will be merged to the submodule when this Pull Request gets merged. |
Background
The so-called agent loop of Auto-GPT currently lacks structural planning and task management. This impedes its long-term performance and also hampers the implementation of retrieval augmentation (see #3536).
Changes & other to-do's
Part 1 - Refactoring (#4799):
agent
->agents
Agent
intoBaseAgent
andAgent
with sensible structurePart 2 - Planning & Task Management (#5077):
Add executor agent (potential solution: specialize existingAgent
implenmentation into an executor agent)Parts / building blocks
The scope and diff of this PR can be magically reduced by merging the PRs below: